HSAB theory and acute metal ion toxicity and detoxification processes

,1. inorg, nucl, Chem, Vol. 40,pp. 2081-2088 © Pergamon Press Ltd.. 1978. Printed in Great Britain

0022-1902/7811201-2081/$02.00/0

Bio-Inorganic Section HSAB THEORY AND ACUTE METAL ION TOXICITY AND DETOXIFICATION PROCESSES MARK M. JONES and WILLIAM K. VAUGHN Departments of Chemistry and Biostatistics and Center in Toxicology,Vanderbilt University, Nashville,TN 37235, U.S.A.

(Received 1 May 1978; received[or publication 8 August 1978) Abslract--The theory of hard and soft acids and bases (HSAB theory) can be used to correlate information on metal ion toxicity and the relative effectiveness of therapeutic chelating agents with parameters used to characterize "hardness" and "softness." HSAB considerations appear to govern at least some of the non-specificaspects of metal ion toxicity. They are also useful in explainingsome of the known patterns of metal ion detoxificationand how situations arise where the rate behavior, rather than equilibriumfactors, may be of importance in determining the success of a given chelating agent in enhancing the excretion of a given toxic metal. In many instances an effective evaluation of the applicabilityof HSAB theory to such systems is prevented by the limited availabilityof the necessary biologicaldata. The correlations found can be used to estimate the toxicity of chemical species for which HSAB parameters but not toxicity data are available, as well as to preselect the most promising chelating agents for therapeutic testing. INTROI~CTION In those types of metal ion toxicity which arise from the metal ion and a generally available donor atom on a sensitive enzyme system (rather than via some highly selective process which can occur with but a single metal ion) one might expect to find some of the same kinds of correlations that are found for other systems where a given coordination site can interact with a reasonably large number of different metal ions. Thus, the correlation of heavy metal toxicities with standard electrode potentials has been demonstrated by Danielli[1]. The correlation of certain types of transition metal toxicity with the Irving-Williams sequence, reported in 1961 by Shaw[2], is apparently of this type. This Irving-Williams order can be expected in many other systems than those reported by Shaw. For example, the log L.D.5o values for + I.O

the intraperitoneal injection of transition metal sulfates into mice also follows this trend as shown in Fig. 113]. Such a correlation, however, can only be used with a limited number of transition metal ions, whose interactions with biological systems do not differ significantly in terms of redox behavior or specificity of interaction. The periodicity of the toxic properties of the metallic elements was clearly demonstrated by Bienvenu et al.[4], who determined the thirty day L.D.5o values for the salts of over forty metallic elements and many non-metallic ones as well, when given to mice intraperitoneally. This periodic correlation with atomic number and the existence of trends within the Jifferent families of elements shown in their results hints strongly at the existence of other correlations more directly connected to various numerical parameters used to

INTRAPERITONEAL, MOUSE LOG L.D.50VALUES VS. ATOMIC NUMBER

•

A

_=

-I.O

7

E

J (9

q

Mn2 .

-2.0

Fe2~"

Co2~"

NiZ÷

I

I

I

I

I

24

25

26

27

28

ATOMIC NUMBER

~÷

Cu

Zn2÷

I

I

29

30

)

Fig. 1. Irving-Williamssequence for the L.D.5ovalues for the divalent transitional metal sulfates, (ip, mice). 2081

2082

MARK M. JONES and WILLIAMK. VAUGHN

characterize the chemical properties of the elements. To search for such correlations of a broader nature we need a more general guiding principle of metal-ligand interactions. Of these more general principles the theory of hard and soft acids and bases (HSAB theory)[5] seems particularly appropriate. HSAB classifies the metal ions into one of three classes: hard (e.g. Bee+, A!3+, Fe 3+, Ca2+, etc.), soft (e.g. Cu+, Ag+, Hg2+, Au+, Pt 2+, etc.) and borderline (Fe :+, Co2+, Ni2+, Zn2+, Pb 2+, Sn2+, etc.). From the view-point of biological donor sites we can say that the hard acids prefer to bind to oxygen or nitrogen, the soft acids prefer sulfur, while the borderline ions form complexes with suffur, oxygen or nitrogen. The borderline ions differ among themselves, of course, with some showing pronounced preferences for oxygen and sulfur over nitrogen (such as Pb 2+ and Sn2+). HSAB theory leads us to expect also that the rates of reactions in which donor pairs are exchanged, will be more rapid for reactions in which the hard/soft character of the products are more effectively matched up. The hard or soft character of the donor species is also important and is rather roughly related to the polarizability of the species involved. For biological systems we are concerned almost exclusively with O, N and S donor types and we can arrange these in order of their increasing softness as O < N < S. The hypothesis that toxicity patterns are related to HSAB concepts was suggested by Pearson several years ago[6]. It was also hinted at more recently by B. Venugopal and T. P. l.,uckey[7] who stated that "the irreversible inactivation of enzymes by metal ions such as C d 2+ and Hg2+ is easily explained by HSAB theory .... " Later in the same section, however, they raised doubts about whether the same types of reactions between metal ion and cellular components as are known to occur in vitro do, in fact, also occur in vivo. They also made no attempt to see if quantitative correlations could be obtained. A direct test for quantitative correlations can be made using available data, which consists of lethality data plus some measure of the hard/soft character of the ions. One example of such a correlation is shown in Fig. 2. Here the logarithms of the L.D.5o values are plotted against the softness parameter trp as tabulated by Ahrland[8]. This parameter is simply the factor:

o'p=

I II

I Acids I/ / //

//Soft //

/ / / eZn2+

/

/

I

3.0

/

/

/ I/re

2+ /

O

.3

//

g

/

/ /

-J

1.0

/,

/

//

I:~ 2.0

II /

eCu

/

/

//

// Go~e~ / /

/ Cd2+ /

/

•

pbZ÷

~2" • ~ / '~,,,Ao / / eA,u + / / / / /I • in 3+

• Snz~

// 1I !

I .050

I .100

I .150

i .200

% Fig. 2. A plot of L.D.50for some ions vs the softness parameter, ~rp. L.D.~o is not a constant, in the sense that most chemical parameters are. For this reason it was considered more realistic to work with L.D.5o dosages plotted logarithmically, thus reflecting the way in which the L.D.~o was calculated in the first place[11]. The units used for L.D.~o values in the literature are generally mglkg. These wereused for preliminary screening, but were transformed to a basis of millimoles/kg for subsequent correlations. The units used for each plot are clearly indicated on that plot. It mu~t also be noted that while acute toxicity probably depends on gross chemical reactivity, chronic toxicity may well have a different basis. As the number and types of metal ions is increased the degree of correlation is altered. The coefficient of correlation of these data is strongly dependent on the

coordinate bond energy of fluoride--coordinate bond energy of iodide coordinate bond energy of fluoride

J

as defined and calculated by Pearson and Mawby[9]. The variety of ion types which one is attempting to include. tabulation of Pearson and Mawby is somewhat more Taking only the points which fall between the dotted extensive than that of Arhiand. Unfortunately, some ions lines, which includes primarily the soft acids, one obtains of considerable interest (e.g. Pd 2+, Pt 2+) are lacking from a fair linear correlation between the L.D.~o values and both tabulations. • the softness parameters, tzp. These data are fit by the The biological parameter selected for use was the equation L.D.5o value. This represents the dosage of a material which will result in an average lethality of 50% of a Log L.D.5o= 26.107trp + 0.0759 population which is large enough to allow usage of the statistical treatment. The problems that arise in the with a correlation coefficient r= 0.9151 and a standard determination of a superficially simple parameter, such deviation of s = 0.256. When additional borderline or as the L.D.5o have been examined in some detail by hard acids are included in the plot shown, the correlation SchLitz[10] and Hayes[ll]. Schiitz provided new experi- is much poorer and the line is represented by the expresmental data on the factors affecting the numerical value sion obtained, as well as a survey of the previous literature. Log L.D.~o= 7.624trp + 1.383, The evidence presented by Schtitz showed that the

HSAB theory and acute metal ion toxicity the correlation coefficient is reduced to r = 0.327 and the standard deviation is increased to s = 0.625. One may conjecture that the correlation is dependent upon differences in the mechanisms of toxicity. From the Figure it is probable that separate treatment of the different kinds of cations might prove more rewarding. For this purpose we have divided our treatment of the data into the same groups used by Pearson to classify the acids, i.e. hard, borderline and soft. The data used in this paper are shown in Table 1. Hard acids. The toxicity of the hard ions may be considered either by groups within the periodic classification or as a whole. Examination of data from a single group shows some of the complications which arise in a clearer fashion than is apparent in a more general approach. If we take the data available on the alkaline earth ions we can see both the striking trends and the extraordinary behavior of ions for which a rather selective mode of toxic action is available. Using data on Mg2+, Ca2+, Sr2+, and Ba 2+, one finds a striking linear correlation between log L.D.5o (mmoles/kg) and the softness parameter irk. This is another one of the softness parameters for metal ions tabulated by Ahrland, that runs from about +5 for the softest ions to - 6 for the hardest. Here the data (Fig. 3) is fit by the line Log L.D.5o = -2.85trk - 5.54

2083

deviation of 0.125, an amazingly good fit. However, adding the data for Be 2+ to this set has a drastic effect on the correlation as would be expected for a metal ion with a different mode of action. For the complete set of alkaline earth metal ions, one finds Log L.D.so = 0.218trk + 1.17 with r = 0.235 and s = 0.747. This shows in a most striking fashion the way in which the correlations can be expected to hold closely only where the mechanisms of the toxic action are identical or very similar. An examination of the data available on 17 metallic ions classified as hard as Pearson, shows that the relationship between the toxicity and the softness parameter is given by the relationship L.D.~o = 1.139 + 7.143tr o with a correlation efficient of 0.501 and a standard error of the estimate of ±0.548 at the 0.02 level of significance. It must be noted that several significant hard ions were eliminated from this correlation because of the unavailability of ~. parameters for them. The cotiimcnt might well be made that such data should be compared on a gram molecular weight basis. The correlation found is then

with a correlation coefficient of -0.988 and a standard

L.D.5o = -1.706+ 11.35o,,

Table 1. L.D.so data used in correlation studies Ion

Salt

Mode of Administration

L.D.50 mg/kg

L.D.50 mmoles/kq

Soft lons Cd2+

CdSO4

i p r mouse

69

.331

Hg2+

HgS04

oral r at

57

.192

Ag2+

AgN03

oral mouse

50

.298

Au+ In 3+

Na3[Au(S203)2] InCl 3

oral mouse

35

.071

ip r mouse

5

.023

T1+ Borderline Ions

TI2SO4

oral mouse

29

.057

Fe2+

FeSO4

ipr mouse

I00

.658

Co2+

CoSO4

ipr mouse

54

.348

Ni 2+

NiSO4

ipr mouse

21

.136

Cu2+

CuSO4

ipr mouse

7

.044

Zn2+

ZnSO4

ipr mouse

29

.180

Pb2+

Pb(OAc)2

ipr mouse

120

.369

SnCI 2

ipr mouse

66

.348

H+

HCl

ip r mouse

40

Li +

LiCI

I p r mouse

604

14.25

Na+

NaCI

Ipr mouse

2602

44.55

K+

KCI

Ip r mouse

552

7.40

Rb+

RbCI

ipr IlK)use

1160

9.59

Cs+

CsCI

~pr mouse

1420

8.09

Be2+

BeCI2

IDr mouse

12

Mg2+

MqCI 2

Ipr mouse

342

3.59

Ca2+

CaCI 2

iDr mouse

280

2.53

Sr 2+

SrCl 2

l p r mouse

908

5.73

Ba2+ A13+

BaCI 2

lpr

mouse

54

.259

ipr

mouse

270

.789

Sc3+

A12(S04) 3 ScCI 3

Ipr

mouse

93

,615

y3+

YCl3

Ipr

mouse

88

,451

La3+

LaCI 3

lpr mouse

121

.694

Cr 3+

CrCI3XH20

lpr mouse

140

.884

Fe3+

FeCI36H20

I p r mouse

250

.962

Sn2+ Hard lons

JINC Vol. 40, No. 12--H

1.097

.150

(Fig. 4)

MARK M. JONES and WILLIAM K. VAUGHN

2084

/ TOXICITY OF ALKALINE. / EARTH IONS AS A FUNCTION/

OF~'K

2.0 r-

/~Mg~

I. O

/ ® Coa+ /

.J

®

a?

0

q

B

~.S

2.0

I.O

-OK

3.0

4.0

Fig. 3. Log L.D.5o (mmole/kg) vs the softness parameter crk for the alkaline earths.

TOXICITY DATA OR HARD METAL IONS

Borderline ions. A plot of the toxicity of the borderline divalent ions, shown in Fig. 5 does show a surprising degree of correlation, considering the diversity of the ions in the group. Here the data can be fit to the curve

2.O • NO÷

LO

O

.;

C

Fe~+ Cr3+"

•

s

.

3+

"H+

Rbt

L.D.5o = 18.79~p - 2.999

+

m. sO÷ / . L , ~ ' e

i

-

d

"

. Boa+ Beg '~

J~" - I.O

-2.0

I O. I00

I

I .200

0350

I

with r := 0.681 and s = 0.304. Here again the correlation is quite striking, with a suggestion that a common mechanism or set of mechanisms may well be operating with the ions in this group. The correlation coefficient here is higher than that found with either hard or soft ions as a class. Soft ions. When we turn to the soft ions, we find a group for which many ~rp values are missing as well as species for which it is difficult to select a characteristic L.D.5o value (i.e. Pt 2+, and even TI+). We are left with the ions Cd 2+, Hg 2+, Ag+~ In 3+ and TI +. There is essentially no correlation of the data as a whole. The data can be "fit" to the line

.2BO

o-p

Fig. 4. Log L.D.5o (mmole/kg) vs the softness parameter ~rp for the hard metal ions. with a correlation coefficient of 0.664 and a standard error of estimate of +0.548 at the 0.25 level of significance. Here again it would be possible to pick Out families of hard ions for which a higher correlation coefficient would be obtained, presumably on the basis of closely related modes of action. Among the "hard!' metal ions there is in fact some considerable variation in the site of greatest sensitivity, a fact due in part to the variable nature of the hydrolytic species actually present at a pH of 7. This has the consequence that one simply cannot examine the toxicity of species such as Th 4+, Zr 4+, U 4+, but only the toxicity of the hydrolyzed forms of each which are stable of blood pH.

Log L.D.5o = -2.75o'p - 0.70, but the correlation coefficient is only -0.366, with a large error of estimate. From the data Fig. 6 we can see that In 3+ and TI+ are significantly more toxic than the other species in this group. If we take only the Group IA and IIA elements the correlation is greatly improved. With these ions the data can be fit to the line Log L.D.~o = 18.76~rp - 1.95 with a correlation coefficient of 0.987 and a standard error of estimate of only 0.059 at the significance level of 0.00628. This correlation is of a special interest as these seem to be metal ions that induce the synthesis of metailothionein[12]. The significantly greater toxicity of the IIIb metals in respect to their ~rp values might lead one to reject TI + as a typical soft cation (its ~p value is much larger than those of the other ions of this group).

HSAB theory and acute metal ion toxicity

2085

0.00 Fe2 + BORDERLINE

IONS

(~

/

J)pb 2~"

-0.60

Zn2"®,~

Ni z+

A

.,?

o Ct

.3

- I . 20

(~Cu2÷

-I .80

¢,9 0

- 2.40

-3.00

L.~;,L

1

0.00

0.05

I

I

0.10

0.15

t

0.20

1

0.25

Fig. 5. Log L.D.5o(mmole/kg)vs the softness parameter o-p for the borderlineions.

We can, however, say that the soft metal ions are species of appreciable toxicity, and the toxicity, except for TI÷, roughly increases with the softness. We see what is probably the same factor operating in those cases where the toxicity of an element is enhanced subsequent to a reduction in oxidation number to a less highly charged, softer ion, e.g. AsS+~ As 3+ and Sn4÷~ Sn2+. A cursory examination of the plots for toxicity does not show the great differences in L.D.~o values between borderline and soft ions that one might anticipate on the basis of their common as well as medical reputations as poisons. Thus the L.D.50 values for poisoning by ZnS04 and CdS04 are not very different. In part this appears to be due to occurrence of different processes being responsible for the toxicity as we pass from one class of acids to the other. Thus the sites of attack and even the enzymes attacked probably vary as we pass from Au+ to Zn 2+ to Be 2+. In part the differences in chronic toxicity are related to the rate of turnover of the metal ion in the mammalian body and the storage sites of the metals involved. Either or both of these may change drastically as we pass from one metal to the next. It must also be remembered that the loss of a metal ion from the mammalian body may correspond to a multicompartment process. Thus a given metal may be removed from the body by a process that corresponds to two or more first order rate processes. This is evidently found with mercury in the data reported by Rothstein

and Hayes[13]. Similarly the kinetics of absorption and excretion of lead in the human body is extremely complex. It is very obvious from the data on soft metal ions that their rate of turnover can be very slow and their halflives for release from certain parts of the body are very long. This may well be due to the lack of naturally occurring serum chelators capable of competing with the biological sites to which they are bound. In this respect the soft metal ions show a significant difference from some hard metal ions whose deposits in the body frequently involve macromolecular hydrolytic species resulting from the hydrolysis of the cation, e.g. Fe 3÷, Al3+, rather than individual metal ions bound very firmly to a set of biological donor sites. General correlation. One may well ask if there is any general correlation of the log L.D.5o with the tro value. Data for twenty-eight ions of all types could be fit to the equation Log L.D.~o = 9.94trp - 1.64 with a correlation coefficient with the surprisingly high value of 0.66 and a standard error of estimate of 0.61 at the 0.00004 level of significance. The conclusion that there is a general correlation between toxicity and softness parameters for a wide variety of metal ions which act as toxins via many routes, does, in fact exist. This

2086

MARK M. JONES and WILLIAM K. VAUGHN

0.00

SOFT IONS -0.40

Cda

LINE FIT TO IONS OF Hg2+( GROUPS IB AND liB ONLY / d

-0.80

E E v oID

\ /

Au + -I.20

(~)Tl+

.J 0 J

-I.60

-2.00 0.00

0.02

0.04

0.06

I 0.08

I 0.10

I 0.12

Fig. 6. Log L.D.5o(mmole/kg)vs the softness parameter ep for some soft metal ions. data is shown in Fig. 7, where the general trend is obvious. Coordination model o f lethal process. One may also ask if there is any model of the overall lethal processes which provides a rationale for the relationship between the incidence of lethality and the softness parameters of the toxic metals. The general biological significance of such discriminating behavior due to selectivity in metalligand interactions has been reviewed by Sigel and McCormick[M]. One such process would be the coordination of the toxic metal to a very sensitive enzyme containing a donor site of some selectivity. The particular enzymes involved would presumably vary from metal to metal. If we assume that the metal and enzyme form a l : l complex which can no longer carry out the essential functions of the enzyme we then have a rough model of the lethal process. This can be represented as

will begin at a higher range of [ME]/[E] values. In fact, one would expect that K could be estimated from the complete lethality curve. These curves are very similar to the curves for the cumulative formation fraction of a complex i.e. a vs log [M]. If we set the [ME]/[E] ratio which corresponds to the L.D.so at X, we can rearrange our expressions as Log g -- Log [ME] - Log [E] - Log [M] or

LOg K = LOg X - Log [lv[]. Now for many sets of complexes log K is inversely proportioned to the softness parameter o-p[15]; so Log.K = -trp + k or

E+M.

Log K o: tro,

" ME then

with Log X - Log [M] ~ - ~ro

[ME] g = [E] [M]"

since the ratio [ME]/[E]--X will be the same at the L.D.so point for the various metals and in general

Toxic symptoms can then be expected to occur when a certain range of [ME]I[E] values is reached and lethality

[M]t = [M] + [ME] ~ [ME],

HSAB theory and acute metal ion toxicity

2087

® 1.60 COMPOSITE DATA CORRELATION

0.80

® ®

®

0.00

®

®

O

E

®®

q,,o _1

®

®

®

- 0.80

O J

® ®

® -

1.60

-2.00

(9 I

I

0.00

I

0.04

0.08

I

I

0.12

i

0.16

0.20

I

0.24

C Fig. 7. Log L.D.~ (mmole/kg) vs the softness parameter ~rp for twenty-eight ions of all categories.

O.IO

.05

:¢ 2-S donors ~,

BAL and Unithiol

== <

==, o

"D Penioillomine

0.15

Cheloting Agentsore nol theropeuticolly ueef~lfor oil of the metalionsshownfoiling in their ronge,but they do tend to form s toble complexeswith most of them/

4.

N, S, 0 donors Co EDTA

0.20

N and O donors

{

~Diethyl Dithiocorbomote,~ S possiblyone N donor

~.~

Softness Parameter

)

Fig. 8. Range of trp values covered by various therapeutic chelating agents.

2088

MARK M. JONES and WILLIAM K. VAUGHN

then [M] = [M], = L.D.5o (mmole/kg) Log L.D.5o (mmole/kg) ~ trp or

Log L.D.5o (mmole/kg) = o-p + constant. Chelating agents. The range of metal ions whose urinary excretion can be enhanced by a given chelating agent is shown in Fig. 8, where the metal ions are placed along a scale of the softness parameter trp. It is apparent that the correlation between ap values and the range of effectiveness of a given chelating agent is not too bad, though there are some obvious imperfections. For example Be 2+ toxicity apparently cannot be very well treated with CaEDTA, though EDTA is capable of enhancing the urinary excretion of the other alkali metals. The literature is also not in accord about the efficiency of CaEDTA in the treatment of maganese poisoning[16]. A very striking demonstration of the operation of the kinetic features of HSAB theory seems to be found in the studies of Hojo, Sugiura and Tanaka[17] who studied the effectiveness of various chelating agents in removing Z°3Hg-labelled organomercury compounds for hemoglobin. They reported that for ligands with the same stability constants for binding the organomercury cations, those which bore sulfur donors were much more effective in removing the organomercury from the hemoglobin! The chart has another use, however, which would be greatly facilitated if more trp values were available. This is the assistance it provides in trying to select a therapeutic chelating agent for a particular metal ion. Thus one would expect that either penicillamine or sodium diethyldithio-carbamate might provide alternative useful antidotes for manganese toxicity. (Please remember, however, that merely removing a metal from the body provides no guarantee that any damage which it has done will heal!) This chart does provide a guide to the formulation of experiments searching for therapeutic chelating agents for metal ions for which none presently exist. In summary then, there appear to be useful correlations between the toxicities of metal ions and parameters measuring their "softness" or "hardness." These correlations can be used as a guide in predicting

the toxicity of chemical species for which such data are not available as well as in the search for therapeutic chelating agents to facilitate the elimination of these from the human body. Acknowledgements--This work was performed under the auspices of the Center of Toxicology, Department of Biochemistry, Vanderbilt University School of Medicine, Nashville, TN 37232 and supported by Grant ES01018-01A1-TOX of the National Institute of Environmental Health Sciences. I also wish to acknowledge with thanks, the assistance provided on the biochemical and toxicological presentation by Dr. Wayland J. Hayes, Jr. and Dr. Thomas M. Harris.

REFERENCES 1. J. F. Danielli, Cell Physiology and Pharmacology, p. 31. Elsevier, Amsterdam 0950). 2. W. H. R. Shaw, Nature 192, 754 (1961). 3. Unless otherwise specified, all toxicity data cited is from The Toxic Substances List 1974 Edition, U.S. Department of Health, Education and Welfare, HEW Publn. No. (NlOSH)74-134, Rockville, MD, 20852 (1974). 4. P. Bienvenu, C. Nofre and A. Cier, Comp. rend. Acad. sci. Paris 256, 1043-44 (1963); C. Nofre, H. Dufuor and A. Cier, Comp. rend. Acad. sci. Paris, 257, 791-794 0963). 5. Hard and soft Acids and Bases (Edited by Ralph G. Pearson). Dowden, Hutchinson & Ross, Pennsylvania (1973); and references therein. 6. R. G. Pearson, Private communication. 7. B. Venugopal and T. P. Luckey, In Heavy Metal Toxicity, Safety and Hormology (Edited by T. P. Luckey), pp. 65--66. Thieme, Stuttgart 0975). 8. S. Ahrland, Structure and Bonding 5, 144 (1968). 9. R. G. Pearson and R. J. Mawby, In Halogen Chemistry (Edited by V. Gutmann), Vol. 3, p. 64. Academic Press, London 0967). 10. E. Schiitz, Arzneimittelforschung 18, 466 (1968). 11. W. J. Hayes, Jr., Toxicology of Pesticides, pp. 39--64. Williams & Wilkins, Baltimore (1975). 12. G. S. Probst, W. F. Bonsquet and T. S. Miya, Toxicology and Applied Pharmacology 39, 61 (1977); and references therein. 13. A. Rothstein and A. D. Hayes, J. Pharm Exptl. Therap. 130, 166 (1960). 14. H. Sigel and D. B. McCormick, Acc. Chem. Res. 3, 201 0970). 15. M. Misono and Y. Saito, Bull. Chem. Soc. Japan 43, 3680 (1970) show log K vs ~p plots for several systems. 16. H. C. Hodge, L. J. Leach, F. A. Smith, W. H. Strain and D. R. Taves, In Drill's Pharmacologyin Medicine (Edited by J. R. Di Palma), 4th Edn, p. 1129. McGraw-Hill, New York (1971). 17. Y. Hojo, Y. Sugiura and H. Tanaka, Radioisotopes 24, 684 (1975).